Anthropic: Calls for a Global Pause on Frontier AI Development

Odaily reported that Anthropic has called on major AI labs to consider slowing down their research and development pace, stating that AI systems are evolving extremely quickly and may soon be capable of self-improvement without human intervention, a capability that could trigger significant social risks. Internal data disclosed by Anthropic shows that the performance of its most advanced models is rapidly improving, and progress in AI models appears to be moving toward "recursive self-improvement," meaning AI systems can autonomously optimize themselves without human intervention. Anthropic stated that it would be beneficial for the world if a global choice could be made to slow down or temporarily halt frontier AI development, allowing social structures and related research to keep pace with technological advancements. Anthropic proposed a global agreement to clearly define how to slow down AI development and establish mechanisms to verify whether competitors are complying with the agreement.